Annals of Botany 78 : 553–558, 1996
Distribution of Cashew Flower Sex-types between Clones and Sides of Tree
Canopies in Tanzania
P. A. L. M A S A W E*†, E. P. C U N D A LL*‡ and P. D. S. C A L I G A RI†§
* Cashew Research Project, Agricultural Research Institute Naliendele P.O. Box 509 Mtwara, Tanzania,
† Department of Agricultural Botany, School of Plant Sciences, The UniŠersity of Reading, Whiteknights,
P.O. Box 221, Reading, RG6 6AS, UK
Received : 27 February 1996
Accepted : 15 April 1996
Production of male, hermaphrodite and abnormal flowers was studied on the north, east, south and west sides of the
cashew tree canopy from Jul. to Dec. 1992 at the Agricultural Research Institute Naliendele, Southern Tanzania.
Flowers were recorded daily on selected clones throughout the main flowering season and subsequently yield of each
clone was also recorded. Results showed differences in the number of flower types both between clones and between
sides ; however, there were consistently more male than other types of flower. Significant yield differences between
clones were not detected and so could not be used to provide unambiguous evidence of the relationship between yield
and flower characteristics but did indicate that clonal differences were present in terms of earliness of yield production.
# 1996 Annals of Botany Company
Key words : Anacardium occidentale L., cashew, flowering, sex-types, clones, Tanzania, sides of canopy, yield.
INTRODUCTION
Nuts of cashew, Anacardium occidentale L., are a high-value
export crop and are the main source of cash income for a
population of more than 2±1 million people in southern and
coastal Tanzania (Shomari, 1988). Little is known of the
factors controlling yield in cashew and in particular the
extent to which yield is influenced by flowering behaviour.
Cashew is reported to be a cross pollinating tree crop
(Pavithran and Ravindranathan, 1974 ; Free and Williams,
1976 ; Palaniswami, Shahul, Hameed and Vijayakumar,
1979). Cashew flowers are borne on an inflorescence which
is an indeterminate panicle. Each flowering panicle possesses
both hermaphrodite and male flowers (Rao and Hassan,
1957 ; Ascenso and Mota, 1972 ; Kumaran, Vimala and
Murthy, 1976 ; Thimmaraju et al., 1980) ; however, abnormal flower types have also been reported (Northwood,
1966 ; Mota, 1973 ; Kumaran et al., 1976 ; Joseph, 1979).
Cashew trees require 4, and even 5, months to complete the
sequential anthesis in the panicle (Pavithran and Ravindranathan, 1974).
The cashew tree normally bears its nuts and attached false
fruit (the cashew ‘ apple ’) on the periphery of the canopy,
and casual observation suggests that one side of the tree
may have a higher nut-set than the other. The formal
existence of such differences has not been established nor
whether the distribution of flower types differs between
sides (e.g. sunny and shaded sides) or whether yield is
directly related to flower distribution. It is also important
for future breeding work, as well as orchard establishment,
‡ Present address : The Forestry Commission, Northern Research
Station, Roslin, Midlothian, EH25 9SY, UK.
§ For correspondence.
0305-7364}96}110553­06 $25.00}0
to determine whether high yield is pre-determined by the
number, distribution in time and}or ratios among the
flower types.
MATERIALS AND METHODS
The experiment was conducted at the experimental station
of the Agricultural Research Institute (ARI) Naliendele,
Mtwara, Southern Tanzania. The Institute is located at a
latitude of 10° 22« S, longitude of 40° 11« E and at an
altitude of 120 m above mean sea level. The mean annual
temperature is around 26 °C and mean annual rainfall is
about 1160 mm which falls in a single 6 month season,
Nov.–Apr. Soil fertility levels are very low and soil structures
are weakly developed. The top soils have loamy sand
textures changing with depth to sandy soil.
Four cashew clones, each represented by four individuals,
vegetatively propagated by air-layering, were selected for
the experiment at ARI Naliendele. The clones studied were
AC4, AC22, AC28 and AZA2 (AC being of Sri Lankan
origin and AZA from Zanzibar). Each clone was planted in
a single row at a spacing of 12 m within and between rows.
For each cashew clone, four trees were taken at random for
the study in the 1992 season. The cashew tree canopy for
each individual selected tree was marked on four sides,
north, south, east and west using a compass. From each
marked side, a total of 15 young panicles of almost the same
size by visual appearance were taken at random for
observation.
Each panicle was tagged and labelled with a panicle
number, giving a total number of panicles for each of the
four clones under observation of 240. The counting of the
opened flowers within each panicle was carried out daily
# 1996 Annals of Botany Company
554
Masawe et al.—Distribution of Cashew Flower Sex-types
T     1. The number of days taken from 30 Jun. 1992 for the first flower to open on different sides of the tree canopy of each
clone
Side
North
South
East
West
Averages
Overall
Types
AC4
AC22
AC28
AZA2
Male
Hermaphrodite
Abnormal
Male
Hermaphrodite
Abnormal
Male
Hermaphrodite
Abnormal
Male
Hermaphrodite
Abnormal
Male
Hermaphrodite
Abnormal
3
3
3
3
5
5
10
18
14
8
11
8
6±0
9±2
7±5
7±58
12
17
15
14
24
18
15
21
18
12
17
15
13±2
19±8
16±5
16±50
25
25
27
27
27
29
34
35
38
25
25
27
27±8
28±0
30±2
28±67
9
20
9
11
9
11
25
32
23
6
12
6
12±8
18±2
12±2
14±42
between 1100 and 1300 h by detaching each from the
cashew panicle using fine watchmaker’s forceps and recording its type. Care was taken to ensure that the residual parts
of the labelled panicles were not physically damaged. A
further three clonal trees of each type studied were selected
for recording yield distribution on the four sides, during the
whole harvesting period (Oct.–Dec.). Powdery mildew
disease was controlled using Bayfidan (Triadimenol 250
EC) at a rate of 30 ml l−" of water.
Three types of flowers were observed and counted
throughout the flowering period. These were male, hermaphrodite and abnormal flowers. Each flower type was
morphologically distinct with male flowers usually having
five sepals, five petals, one large exserted stamen (occasionally two) and 5–16 small inserted stamens, each stamen
comprising an anther and a short filament. The large stamen
is nearly twice the length of the small stamens. The large
stamen and most of the small stamens produce pollen.
Hermaphrodite flowers were rather similar to the male
flowers but had an additional flower part, the pistil, which
was normally longer than the large stamen but occasionally
was shorter or of equal size. In this experiment ‘ abnormal
flowers ’ were taken to include all flowers whose staminal
arrangement was similar to that of both male and
hermaphrodite flowers but having neither pistil nor large
stamen. Small stamens might have been present but were
not counted.
Analyses of variance were carried out on the data collected
using SAS (SAS Institute, North Carolina, USA), in three
parts : (a) numbers of flower types produced between 1 Jul.
and 30 Sep. (the period considered to be the most
commercially important), (b) the number of flower types at
the highest peak of production regardless of the time and,
(c) the total number of flower types from 1 Jul. to 30 Dec.
Types
Average
Sides
Average
12±25
16±25
13±50
13±75
16±25
15±75
21±00
26±50
23±25
12±75
16±25
14±00
14±94
18±81
16±62
14±00
15±25
23±58
14±33
RESULTS
Production of flower types
‘ Flowering day ’ was taken as the number of days, from 30
Jun. 1992, for the appearance of the first flower of each type
(male, hermaphrodite or abnormal) to open. Thus 1 Jul.
was counted as day 1, and so on. Flowering day varied
between cashew clones (Table 1). Amongst the clones, AC4
was clearly the earliest, producing the first flowers on day 3
on the north side, together with male flowers on the south
side. Clone AZA2 was the next earliest followed closely by
AC22. Clone AC28 was considerably later than the other
three clones, taking nearly a month. When attention is
turned to sides of the trees, the north side produces flowers
first, on average 14 d, followed very closely by the west side
(14±3 d), the south (15±3 d) and later by the east side at
23±6 d. In terms of the different flower types all clones first
produced male flowers followed by abnormal ones and
hermaphrodite ones.
It would thus seem that at this site, the north and west
sides of the tree favour early flowering. The production of
all flower types increased with time as can be seen from Fig.
1 A–D. The figures show only the number of flowers on 15
panicles for the north side of each clone but all other sides
showed a similar pattern. However, as can be seen the
production of male flowers increased dramatically as
compared to hermaphrodite and abnormal flowers. The
trend of production of male flowers was quite similar in all
clones, with one distinct peak and one more subtle, later
peak. The major production of male flowers was more
pronounced on clone AC22 (Fig. 1 A) but there tended to be
more hermaphrodite flowers produced by clone AC28 (Fig.
1 B). On the other hand, clone AZA2 was the most prolific
producer of abnormal flowers (Fig. 1 C), followed by clone
AC4 (Fig. 1 D). Hermaphrodite flowers had a different
pattern of production from that of males with a broad
spread over time in all clones. The number of hermaphrodite
120
A
100
80
60
40
20
0
28 Jun. 26 Jul. 23 Aug. 20 Sep. 18 Oct. 15 Nov. 13 Dec.
12 Jul. 09 Aug. 06 Sep. 04 Oct. 01 Nov. 29 Nov. 27 Dec
Number of flowers/15 panicles (Counts)
Number of flowers/15 panicles (Counts)
Masawe et al.—Distribution of Cashew Flower Sex-types
120
C
100
80
60
40
20
0
28 Jun. 26 Jul. 23 Aug. 20 Sep. 18 Oct. 15 Nov. 13 Dec.
12 Jul. 09 Aug. 06 Sep. 04 Oct. 01 Nov. 29 Nov. 27 Dec.
Dates
B
100
80
60
40
20
0
28 Jun. 26 Jul. 23 Aug. 20 Sep. 18 Oct. 15 Nov. 13 Dec.
12 Jul. 09 Aug. 06 Sep. 04 Oct. 01 Nov. 29 Nov. 27 Dec
Number of flowers/15 panicles (Counts)
Number of flowers/15 panicles (Counts)
Dates
120
555
120
D
100
80
60
40
20
0
28 Jun. 26 Jul. 23 Aug. 20 Sep. 18 Oct. 15 Nov. 13 Dec.
12 Jul. 09 Aug. 06 Sep. 04 Oct. 01 Nov. 29 Nov. 27 Dec.
Dates
Dates
F. 1. The numbers of male (*), hermaphrodite (8) and abnormal flowers (+), per 15 panicles, recorded over time (28 Jun. to 28 Dec. 1992)
on the north side of the four clones A, AC22 ; B, AC28 ; C, AZA2 and D AC4.
T     2. The total number of flowers counted on 60 panicles ( from four trees) of indiŠidual clones for different sides of the
tree canopy oŠer the 1992 season
Type
Side
Male
North
South
East
West
Total
Percent
North
South
East
West
Total
Percent
North
South
East
West
Total
Percent
Hermaphrodite
Abnormal
Total flowers
Mean number of flowers per panicle
AC4
9 762
6 754
6 371
7 746
30 633
58±2
5 180
4 642
4 145
5 114
19 081
36±3
829
765
554
741
2 889
5±5
52 603
219
AC22
AC28
AZA2
Mean
20 467
18 329
15 161
16 105
70 062
76±7
4 001
5 681
5 275
5 081
20 038
21±9
517
504
526
447
1 994
1±3
92 094
384
13 733
13 098
11 220
14 655
52 706
57±7
8 806
8 560
9 374
10 710
37 450
41±0
325
269
245
280
1 119
1±2
91 275
380
10 533
9 806
7 370
13 185
40 894
53±8
6 297
4 641
5 039
7 296
23 273
30±6
821
658
572
900
2 951
15±5
67 118
280
13 624
11 997
10 031
12 923
48 575
64±1
6 071
5 881
5 958
7 050
24 960
32±9
623
549
474
592
2 238
2±9
75 773
316
556
Masawe et al.—Distribution of Cashew Flower Sex-types
T     3. Analyses of Šariance for production of flower types
of the four clones oŠer the four sides of the tree
production of male flowers that were traceable to the sides
of the tree, while the interaction of these effects was used as
the error term.
The percentages of male, hermaphrodite and abnormal
flowers were variable between clones (Table 2). Male flowers
were always the most frequent and ranged between 53±8 %
(AZA2 clone) to 76±7 % (AC22 clone) while hermaphrodite
flowers had a lower frequency and were between 21±9 %
(AC22 clone) and 41±0 % (AC28 clone). It was interesting to
note a high percentage (15±5 %) of abnormal flowers in the
clone AZA2, as opposed to clone AC28 which only displayed
1±2 %.
The ratio of male to hermaphrodite flowers can be
calculated from the data in Table 2 and showed there was
variability between clones and sides ; ranging from 1 : 1±19
for the east side of clone AC28 to 1 : 5±12 for the north side
of clone AC22. However, considering the overall number of
flowers, summing over sides, clone AC22 stood out as
having a high ratio, 1 : 3±50, while the others all had
moderate ratios with an excess of males of 1±41, 1±59 and
1±76. Although individual sides had different ratios within a
clone there was general agreement in the relative size of the
ratios.
The distributions of the yield (g) over time for the four
clones are presented in Fig. 2. It is clear that clone AC28
was relatively late yielding as compared to others. The yield
on different sides of the trees, alongside the number of
hermaphrodite flowers, is presented in Table 4. There was
no obvious correlation between yield and numbers of
hermaphrodite flowers. However, we should note that the
analysis of variance (Table 5) showed no significant
Mean square
Item
d.f.
Male
Hermaphrodite
Abnormal
Clone
Side
Error
3
3
9
1 700 499***
686 085***
85 612
299 959***
44 795
26 953
204 645***
6 037
3 286
*** P! 0±001.
flowers was relatively low compared with the number of
male flowers and there were no sharp peaks of production.
The total number of each type of flower varied from clone
to clone as shown in Table 2. The count was based on 60
panicles per side per clone, being 15 per side on each of four
trees for each clone. Clone AC4, although it was the earliest
flowering actually produced the fewest flowers (52 603,
equivalent to an average of 219 per panicle) in total, and was
lowest for both numbers of male and hermaphrodite flowers.
Clones AC28 and AZA2 produced the highest totals, 91 275
and 92 094 respectively (380 and 384 flowers per panicle).
Results of the analysis of variance for the production of
the flower types using the different parts of the distribution
gave similar levels of significance and patterns of variability.
Therefore, only the results for flower production between 1
Jul. and 30 Sep. are shown in Table 3. As can be seen, for
all flower types there are significant effects attributable to
clonal differences. There were significant effects on the
1100
1000
900
800
Yield (g)
700
600
500
400
300
200
100
0
21 Sep.
05 Oct.
28 Sep.
19 Oct.
12 Oct.
02 Nov.
26 Oct.
16 Nov.
09 Nov.
30 Nov.
23 Nov.
14 Dec.
07 Dec.
28 Dec.
21 Dec.
F. 2. The yield (g) distribution over time of the four clones in the 1992 season. ( ) Clone AC22 ; (——) clone AC28 ; ( * ) clone A2A2 ; ( ^ )
clone AC4.
Masawe et al.—Distribution of Cashew Flower Sex-types
T     4. Comparison of mean number of hermaphrodite
flowers and yield of nuts (g) on different sides of the tree
canopy of selected clones
Sides of the cashew tree canopy
Clone
Flower}yield
North
South
East
West
Total
AC4
Number
Yield
Number
Yield
Number
Yield
Number
Yield
Number
1295
1541
1000
1287
2201
2269
1574
3355
1517
2113
1160
1217
1420
960
2140
1669
1160
1841
1470
1422
1036
1677
1319
1115
2343
1727
1260
2443
1489
1741
1278
1556
1270
1124
2677
2316
1824
2074
1762
1767
4769
5991
5009
4486
9361
7981
5818
9713
—
—
AC22
AC28
AZA2
Mean
Yield
T     5. Analysis of Šariance for the yield (g) of indiŠidual
clones by side
Source
d.f.
Mean Square
F-value
P
Clone
Side
Clone¬Side
Error
3
3
9
32
44±8
85±5
9±1
139±1
0±32
0±61
0±07
NS
NS
NS
differences between sides of the tree or between clones for
yield (transformed to log ).
"!
DISCUSSION
There were differences between clones in the date of first
flower opening and in the time and duration of peak
flowering. There is, therefore, the possibility of carrying out
selection for earliness of flowering as well as the duration of
flowering. This latter characteristic is important, as extended
flowering may lead to undesirably late nut}fruit production,
coinciding with early onset of rains and conflicting labour
requirements with other crops (Northwood, 1966). Some
clones, such as AC4, AC22 and AZA2, peaked early and
yielded over a short period while others, like AC28, yielded
over a wider spread of time.
The mean number of flowers per panicle was variable,
ranging from 219 (clone AC4) to 384 (clone AC22) which
are inside the range reported by other authors. In India,
Damodaran (1966) observed 486 flowers per healthy panicle
while Hanamashetti et al. (1986) reported a range from 165
to 837. Heard, Vithanage and Chako (1990) observed 16
panicles over 50 d in Australia and noted a mean number of
443 flowers per panicle.
In most cases the first flowers to open were male, as was
reported by Moranda (1941), Rao and Hassan (1957),
Northwood (1966) as well as Pavithran and Ravindranathan
(1974). For most of the season male, hermaphrodite and
abnormal flowers were all open at the same time, but the
number of male flowers was considerably greater than the
number of hermaphrodite and abnormal flowers. The
557
production of abnormal flowers was very low in clones
AC28 and AC22, but appreciable for clones AZA2 and
AC4.
There were highly significant differences in abundance of
male flowers between clones, and between the sides of the
same clone. By contrast, the differences in numbers of
hermaphrodite flowers varied significantly between clones
but there were no significant differences between sides.
The ratio of male to hermaphrodite flowers varied between
the clones and the different sides of the same clone. In most
clones higher ratios were found on the north side.
Considering that Tanzania is to the south of the equator
while India is to the north, the present results agree with
those reported by Chakraborty, Sadhu and Bose (1981).
These latter authors found that the panicles on the south
side gave a maximum number of hermaphrodites and a
higher ratio of male to hermaphrodite flowers, while the east
side gave the lowest. They suggested that the distribution of
flowers was influenced by light and temperature.
It has been claimed by Wunnachit and Sedgeley (1992)
that the number of hermaphrodite flowers can be used as a
selection criterion, and Heard et al. (1990) reported that
pollination was not a limiting factor for cashew production.
In the present study the yields of cashew in 1992 were
generally low and it was therefore not surprising that there
were no significant yield differences detected between clones
or between sides of the same clone which could be related to
pattern of flowering. However, it is noteworthy that the
north side (the ‘ sun-facing ’ side) did give the highest yields.
There was a continuous production of hermaphrodite
flowers from the beginning of flowering to the end of
November while the production of male flowers decreased
with time. In parallel with this it was noted that the
hermaphrodite flowers produced in November produced
few or no nuts. Thus indicating the potential importance of
the male flowers, rather than hermaphrodites, in determining
yield potential as was suggested to be the case in the
Philippines (Moranda, 1941).
Cashew flowers are self-fertile, but probably not self
pollinating, as bagged flowers set no fruit (Northwood,
1966 ; Free and Williams, 1976 ; Masawe, 1994) but selfpollination by hand improved fruit set (Northwood, 1966).
The presence of scent, nectar, coloured petals and sticky
pollen all suggest insect pollination. The relative importance
of insect and wind pollination has, however, not been
critically established. It would also add considerable
confidence to devising future strategies if further more
extensive research was carried out on yield performance on
the different sides of the cashew tree across a range of
locations. Nevertheless, the present results agree with earlier
reports and re-enforce the potential that selection for floral
behaviour could give beneficial results for cashew production.
A C K N O W L E D G E M E N TS
Financial support from the Tanzanian Government and the
British Government through ODA is gratefully acknowledged. We also wish to acknowledge the help of the
technical staff of the Breeding Section, ARI Naliendele for
their assistance in the experimental work.
558
Masawe et al.—Distribution of Cashew Flower Sex-types
LITERATURE CITED
Ascenso JC, Mota IM. 1972. Studies on the flower morphology of
cashew (Anacardium occidentale L.). Agronomia Moçambicana
(Mozambique) 6 : 107–118.
Chakraborty DK, Sadhu MS, Bose TK. 1981. Studies on sex expression
in cashew (Anacardium occidentale L.). ProgressiŠe Horticulture
13 : 1–3.
Damodaran, VK. 1966. The morphology and biology of the cashew
flower, Anacardium occidentale L. II. Anthesis, dehiscence,
receptivity of stigma, pollination, fruit set and fruit development.
Agricultural Journal of Kerala (India) 4 : 78–84.
Free JB, Williams IH. 1976. Insect pollination of Anacardium
occidentale L., Mangifera indica L., Blighia sapida Koeng and
Persea american Mill. Tropical Agriculture 53 : 125–139.
Hanamashetti SI, Khan MM, Mahabaleshwar H, Mallik B, Sulladmath
UV. 1986. Flowering and sex ratio in some cashew (Anacardium
occidentale L) selections. Journal of Plantation Crops 14 : 1, 68–70.
Heard TA, Vithanage V, Chako EK. 1990. Pollination biology of
cashew in the northern territory of Australia. Australian Journal of
Agricultural Research 41 : 1101–1114.
Joseph KT. 1979. The cashew flower. Indian Cashew Journal 12 : 7–9.
Kumaran PM, Vimala B, Murthy KN. 1976. On occurrence of pistillate
and neutral flowers in cashew. Journal of Plantation Crops 4 :
82–84.
Masawe PAL. 1994. Aspects of breeding and selecting improŠed cashew
genotypes. PhD Thesis, The University of Reading.
Moranda EK. 1941. Cashew culture. Philippine Journal of Agriculture
(Philippines) 12 : 89–106.
Mota MI. 1973. Flower abnormalities in cashew (Anacardium
occidentale L.). Agronomia Mocambicana 7 : 21–35.
Northwood PJ. 1966. Some observations on flowering and fruit-setting
in the cashew, Anacardium occidentale L. Tropical Agriculture
(Trinidad and Tobago) 43 : 35–42.
Palaniswami V, Shahul Hameed A, Vijayakumar M, 1979. Vegetative
propagation in cashew-work done at Vridhachalam. In : Bhaskara
Rao EVV, Hameed Khan H, eds. Proceedings of the International
Cashew Symposium, Kerala, India : Acta Horticulturae, 67–70.
Pavithran K, Ravindranathan PP. 1974. Studies on floral biology in
cashew, (Anacardium occidentale L.). Journal of Plantation Crops
2 : 32–33.
Rao VNM, Hassan MV. 1957. Preliminary studies on the floral biology
of cashew (Anacardium occidentale L.). Indian Journal of Agricultural Science (India) 27 : 277–288.
Shomari SH. 1988. A review on cashew research in Tanzania. In :
Tanzania Agriculture Research Masterplan Conference, Arusha.
Thimmaraju KR, Narayana Reddy MA, Suryanarayana Reddy BG,
Sulladmath UV. 1980. Studies on the floral biology of cashew
(Anacardium occidentale L.). Mysore Journal of Agricultural
Science 14 : 490–497.
Wunnachit W, Sedgley M. 1992. Floral structure and phenology of
cashew in relation to yield. Journal of Horticultural Sciences 67 :
769–777.